Nirvana: Clinical Grade Variant Annotator

Sequencing an individual genome typically produces approximately three million variants compared to the human reference genome. The consequence for each variant depends on the location and nature of the variant and is a key question for genetic analysts performing clinical diagnosis. Variant annotation describes how a variant affects the sample's genome. These annotations include the functional consequence on the different transcripts for a gene or in proximal regulatory regions. Annotation also includes additional data on what is known about a given variant that can help in understanding its relevance to a given line of investigation. Often this data is provided by different sources and contain allele frequencies for different populations, clinical implications, relevance to cancer types, additional studies, etc. Ultimately this information helps clinicians interpret variants when providing a diagnosis. The three most widely used open source annotation tools are VEP, SnpEff and AnnoVar. VEP is widely considered to be most accurate of the three, but is also slower than both SnpEff and AnnoVar. When annotating the variants from a 30x genome (NA12878), VEP finished in 18 hours whereas SnpEff 4.3g and AnnoVar finish in 15 min and 67 min respectively using one core. We present Nirvana, an open source clinical variant annotator, that is both accurate (over 99.9% concordance with VEP) and fast (takes 7 min to annotate NA12878). Nirvana is used in all of Illumina's relevant analysis pipelines and is tested rigorously to ensure adherence to clinical standards.